|
|
A formal approach to understanding redundancy
Hedde Zeijlstra University of Amsterdam Thursday, February 19, 10:30AM, MJH Rm 126 Natural language exhibits abundant usage of redundant markers that have to be obligatorily present. Examples are subject-verb agreement (1), negative concord (2), modal concord (3) and sequence of tense (4). In all these cases there is a second marker of a particular semantic function of which it us not clear how it contributes to the meaning of the sentence, if it contributes to it at all. (1) John walks (2) I ain't seen nothing (AAVE) (3) You can't possibly win that game (4) John said Mary and Bill looked ill Most concord phenomena have been given different analyses. For instance for negative concord several analyses have been provided: De Swart & Sag (2002) take it to result from resumptive quantification, in Giannakidou (2000) it has been attempted to take n-words in negative concord languages to be NPIs and in Zeijlstra (2004, 2008), Negative Concord is taken to be an instance of syntactic agreement, not principally different from the phenomenon illustrated in (1). In this talk I argue that mechanisms such as the one provided for Negative Concord in Zeijlstra (2004, 2008) may also underlie the other phenomena addressed above. In short, it is argued that all these instances of doubling are the result of a (multiple) agreement relation between one single interpretable feature and one or more uninterpretable features. But if the following hypothesis is true, a new question immediately arises: why would natural language exhibit uninterpretable features in the first place? In this talk I provide a formal motivation for the existence of uninterpretable features: it allows natural language to spell out the presence of a particular semantic operator on another lexical item. 
|